The Dawn Horse Is No Longer the Ancestor of Horses

The mammalian order Perissodactyla includes today's horses, tapirs, and rhinoceroses. Yet during the Paleogene, these animals were far more diverse and widespread across the Northern Hemisphere. The puzzle lies in their sudden appearance in the fossil record: around the time of the Paleocene–Eocene Thermal Maximum (PETM), about 56 million years ago, perissodactyls emerged almost simultaneously in North America, Europe, and Asia, and they already displayed considerable diversity. Compared with artiodactyls and other mammalian groups, the early radiation of perissodactyls remains unusually enigmatic. Because modern perissodactyl diversity has declined drastically, reconstructing their evolutionary history using DNA is extremely difficult. As a result, most phylogenetic reconstructions rely primarily on morphological evidence derived from fossils and skeletal material of living species.

A study published in PNAS in 2026 focused on early Eocene perissodactyl species and constructed a morphological dataset including 71 taxa and 101 characters. The researchers conducted a phylogenetic analysis using maximum parsimony. In their matrix, they also incorporated intraspecific polymorphism and treated many characters as additive multistate characters. The analysis produced 1,254 most parsimonious trees, and the resulting 50% majority-rule consensus tree revealed that many species traditionally regarded as ‘early horses’ are not actually members of the true horse lineage (Hippomorpha).

Phylogenetic tree of Perissodactyla, with the position of Hyracotherium marked by a red star（圖片來源：Tissier J et al. (2026)，採用 CC BY-NC-ND 4.0 授權）

Among these taxa, the genus Hyracotherium, long celebrated as the earliest ancestor of horses, turns out not to be a true horse at all. The species Hyracotherium leporinum itself, along with several North American species historically assigned to the genus, should be excluded from the hippomorph lineage. Consequently, the familiar textbook illustration showing a linear evolutionary progression from Hyracotherium directly to modern horses no longer matches current phylogenetic evidence.

Within the revised evolutionary framework, the earliest genuine members of the hippomorph lineage appear slightly later in the early Eocene Wasatchian stage. North American genera such as Systemodon and “Xenicohippus” represent some of the earliest branches of this group. This suggests that the true origin of the horse lineage occurred somewhat later than previously assumed. Earlier small mammals with relatively simple dentitions—once interpreted as ancestral horses—are now regarded as stem perissodactyls rather than members of the horse lineage.

Another important revision concerns the classification of several genera. The genera Sifrhippus and Arenahippus, previously considered distinct early equids, are now interpreted as junior synonyms of Pliolophus. However, Pliolophus itself does not belong to the hippomorph lineage either and is instead classified as a stem perissodactyl. Similarly, the genus Chowliia, established in 2006, is now regarded as a junior synonym of Cardiolophus. Although Cardiolophus had traditionally been assigned to the tapiromorph lineage, the new phylogenetic analysis places it among basal perissodactyls. These taxonomic revisions simplify the early evolutionary picture of perissodactyls, suggesting that previous studies may have overestimated the number of genera by splitting taxa based on minor dental differences.

Morphologically, Hyracotherium, Cardiolophus, and Pliolophus can be distinguished by several cranial and dental features. These include the position of the orbit and infraorbital foramen, the presence or absence of a diastema between premolars and canines, and the structure of crests on the upper and lower molars. Detailed comparisons of these features reveal that early perissodactyls comprised multiple distinct evolutionary branches rather than a single linear lineage.

At a higher taxonomic level, the study supports the division of crown-group perissodactyls into two major clades: Hippomorpha and Tapiromorpha. Hippomorphs share several diagnostic traits, such as the orientation of the cristid obliqua on lower molars toward the metaconid. Tapiromorphs, in contrast, differ in features such as the morphology of the paraconule on upper molars and the length–width proportions of molar teeth. The family Chalicotheriidae, long debated in perissodactyl systematics, is placed within Tapiromorpha in this analysis.

In terms of paleobiogeography, the phylogenetic reconstruction suggests that perissodactyls may have originated in the Indian subcontinent or nearby parts of Asia. The most basal taxon in the analysis, Gazijhippus, comes from Pakistan, although its age is slightly younger than the earliest known Northern Hemisphere perissodactyls. This discrepancy implies the existence of ghost lineages—evolutionary branches for which fossil evidence has not yet been discovered. Other taxa closely related to perissodactyls, such as Perissobune and Cambaytherium, also come from the India–Pakistan region, further supporting this geographic hypothesis.

Overall, the newly reconstructed evolutionary tree substantially reshapes our understanding of early perissodactyl history. It redefines the evolutionary placement of the so-called dawn horse and reveals a far more complex pattern of diversification within the early perissodactyl lineage.

Note: An additive multistate character describes a trait with three or more states arranged in a logical sequence, such as 0 = low-crowned teeth, 1 = moderately crowned teeth, and 2 = high-crowned teeth. In contrast, a non-additive multistate character lacks such an ordered progression, for example 0 = red, 1 = blue, and 2 = green.

Author: You-Ye Shui

Reference:

Tissier J et al. (2026). Earliest perissodactyls reveal large-scale dispersals during the PETM. PNAS.